Searching 100M images by content similarity

In this paper we present the web user interface of a scalable and distributed system for image retrieval based on visual features and annotated text, developed in the context of the SAPIR project. Its architecture makes use of Peer-to-Peer networks to achieve scalability and eciency allowing the management of huge amount of data and simultaneous access by a large number of users. Describing the SAPIR web user interface we want to encourage nal users to use SAPIR to search by content similarity, together with the usual text search, on a large image collection (100 million images crawled from Flickr) with realistic response time. On the ground of the statistics collected, it will be possible, for the rst time, to study the user behavior (e.g., the way they combine text and image content search) in this new realistic environment.